Innovations to consumer confidence convey incremental information about economic activity far into the future. Does this reflect a causal effect of animal spirits on economic activity, or news about exogenous future productivity received by consumers? Using indirect inference, we study the impulse responses to confidence innovations in conjunction with an appropriately augmented New Keynesian model. While news, animal spirits, and pure noise all contribute to confidence innovations, the relationship between confidence and subsequent activity is almost entirely reflective of the news component. Confidence innovations are well characterized as noisy measures of changes in expected productivity growth over a relatively long horizon. (JEL D12, D83, D84, E12)

Abstract The basic New Keynesian model predicts that positive supply shocks are less expansionary at the zero lower bound (ZLB) compared to periods of active monetary policy. We test this prediction empirically using Fernald's (2014) utilization-adjusted total factor productivity series, which we take as a measure of exogenous productivity. In contrast to the predictions of the model, positive productivity shocks are estimated to be more expansionary at the ZLB compared to normal times. We find that there is no significant difference in the response of expected inflation to a productivity shock at the ZLB compared to normal times.

Abstract This paper develops a New Keynesian model featuring financial intermediation, short- and long-term bonds, credit shocks, and scope for unconventional monetary policy. The log-linearized model reduces to four equations: Phillips and IS curves, as well as policy rules for the short-term interest rate and the central bank's long-bond portfolio (QE). Credit shocks and QE appear in both the IS and Phillips curves. In equilibrium, optimal monetary policy entails adjusting the short-term interest rate to offset natural rate shocks but using QE to offset credit market disruptions. Use of QE significantly mitigates the costs of a binding zero lower bound.